China Securities Co., Ltd.: The moment of AI education breakthrough is coming, empowering teaching and research to improve quality and efficiency.

China Securities Co., Ltd. released a research report stating that ChatGpt has ushered in the era of generative AI, while DeepSeek has significantly reduced costs, and a transformation in AI education is imminent. China Securities Co., Ltd. believes that AI "substituting tutoring teachers" will be much easier than "substituting lecturers," and in the future, every student may need an AI learning assistant. Currently, AI education is mature in areas such as photo-based question search, error correction, assessments, and personalized tutoring, empowering enterprises to improve quality and efficiency. However, there is room for improvement in precise learning. The greatest potential application of AI education is for extracurricular learning and assessment. The commercial directions include: 1) intelligent education hardware supported by large models, integrating Q&A, assistance, and teaching functions, with the potential to replace the functions of tutoring teachers (answering questions, grading, assessments, practice processes) and some teaching teachers (knowledge explanation, learning processes); 2) AI education 2C software applications tailored to specific educational scenarios (language learning, Q&A); 3) B-end application scenarios. The main viewpoints of China Securities Co., Ltd. are as follows: Why do we believe that AI is at a critical point in transforming education? Learning is a slow process and often has traits that are counterintuitive and present a "impossible triangle" (personalization, high quality, large scale/low cost). However, each technological innovation cycle has brought significant changes to the education industry, such as the emergence of online large class webinars and question-answer apps in the era of mobile internet. ChatGpt's generative AI has ushered in a new era of AI with larger scale models achieving good interactive Q&A, multimodal understanding, and generation of various types of content abilities. Released in January 2025, DeepSeek has low costs, open-source capabilities, and reasoning abilities, which will improve the precision of AI education, promote deep thinking, and save training costs. AI-enabled heuristic teaching will become possible. With the continuous evolution of large models, a new transformation in AI education is on the horizon, and it is expected to accelerate the emergence of education-native apps in the era of large models. Where should AI education start? What are the commercial scenarios? AI's empowerment in education scenarios can be divided into teaching, learning/practice/assessment, management, and examinations. Currently, in terms of maturity, AI education is relatively mature in areas such as photo-based question search, error correction, assessments, and personalized tutoring, and can improve teaching and research to enhance quality and efficiency. However, there is still room for improvement in precise learning. We believe that in AI's transformation of education, it is easier to "substitute tutoring teachers" than "substitute lecturers," and AI learning assistants may become a possibility. The commercial directions include: 1) intelligent education hardware supported by large models, integrating Q&A, assistance, and teaching functions, which has the potential to replace the functions of tutoring teachers and some teaching teachers, in the post-double reduction era, the sales volume of learning machines (including practice machines) in 2024 exceeded 6 million units, a growth of 25.5% (with online sales accounting for over 90%), accounting for approximately 3% of in-school K12 students. Assuming a future penetration rate of over 30%, this corresponds to over 50 million units, and under the support of AI, learning/practicing machines are expected to achieve a transformation from 1 to 10 in the industry. 2) AI education 2C software applications tailored to specific educational scenarios, such as language learning and Q&A. 3) B-end application scenarios to further enhance the efficiency of teachers' work.